[Grub-dev] GrubNG C# client 0.5 released
Balinny
balinny at gmail.com
Wed Apr 16 20:27:22 UTC 2008
Yousef Ourabi wrote:
> Most of what I think is common to *every* client is what you have in
> your "core" module -- mostly to do with parsing arc and re-creating
> the work-unit to upload -- since every port will have to do this, and
> it's essentially the same no matter what it might be best to pull out
> into C bindings for different languages.
>
> More soon, and sorry fro my delayed reply.
>
By parsing and recreating workunits i suppose you mean when the work is
stopped? How is the csharp managing upload resuming?
The reasons the C client recrawls teh workunit instead of resuming the
files are:
a) It isn't that expensive at the end of the day crawling a workunit
instead of a half.
b) It is easier to restart when things go wrong. I.e. the network goes
down and every url fails without being able to resolve the dns. When
restarting after the problems are gone, the workunit will be recrawled,
so you don't have those urls skipped.
c) It's easier to directly create the crawled data gzipped. I don't know
how to resume a gz file without producing decompression error.
Perhaps i should make it create a workunit importing from a different
data file successful crawls.
More information about the Grub-dev
mailing list